A learning health system for people with severe mental illness: a promise for continuous learning, patient coproduction and more effective care

Abstract

A Learning Health System (LHS) promotes the patient being at the very center of his or her care. Patient coproduction of care in an LHS is enabled by a focus on improving outcomes through the use of tools and visualizations that use the harnessed knowledge obtained from every previous treatment of similar patients. Interest in the concept of LHS is growing, and there are promising results in real-world applications. Almost no research has focused on LHSs for severe mental illness (LHS4SMI). By using a user-centered system design approach, a persona and use-case scenarios were created to illustrate how schizophrenia care could be co-produced in an LHS compared to standard care in a non-LHS. The illustration highlight increased participation through decisions informed by all treatments for all similar patients through the use of user interfaces that support continuous evaluation, increased understanding, compensation for cognitive impairment and participation of next of kin in the care process. We propose that an LHS4SMIs like schizophrenia has enormous potential in enabling continuous learning, patient coproduction, and more effective care.

The concept of Learning Health Systems (LHS)

A Learning Health System (LHS) is defined as, ‘one in which science, informatics, incentives, and culture are aligned for continuous improvement and innovation, with best practices seamlessly embedded in the care process, patients and families active participants in all elements, and new knowledge captured as an integral by-product of the care experience.’ [1, p.137]. The concept has been developed for more than a decade by experts from a wide range of disciplines, e.g. clinical researchers, clinicians, healthcare analytics, statisticians, biostatisticians, epidemiologists, healthcare informaticians, research funders, health products industry, payers, and regulators, to find a way to draw on the strengths of each domain [2].

The patient at the center

At the very center of an LHS is the patient, the coproduction of care and joint evaluation of outcomes [1]. Coproduction refers to the ‘interdependent work of users and professionals who are creating, designing, producing, delivering, assessing, and evaluating the relationships and actions that contribute to the health of individuals and populations.’ [3, p.2]. By focusing on improvement of the outcomes that matter the most to patients, data is collected, analyzed and presented in formats that enable shared decision making (SDM) [4,5]. Visualizations of progress and prognostic information in an LHS can not only support patients in coproducing care but enable continuous improvement and learning with the acquisition of data. ‘With sufficient aggregated data, this system could be used to develop powerful outcome prediction tools adapted to the unique characteristics of the user and better personalize the delivery of healthcare.’ [6,p.7]. Whether through patient portals accessed by the individual or data visualizations and clinical decision support systems (CDSS) accessed during visits, the latest knowledge generated through all data within the system is accessible in useful formats supporting personalized care.

A sociotechnical system enabling multiple uses of the same data – better individual care, quality improvement (QI) and research

The LHS is a socio-technical endeavor in that it focuses on the development and joint optimization of the social part of the system (work processes) and supportive technologies [7]. The technical side is primarily a platform that allows collection, real-time analysis, and visualization of data for a broad range of purposes [8]. It is used, e.g. creating a real-time evidence-base for decision-making using all available cases with similar properties, making a predictive analysis on prognosis and risks and applying intelligent automation to lessen administrative tasks [9]. Emphasis is put on data visualizations adapted to users on all different levels in the organization, generating learning through feedback. Data is analyzed and used in visualizations not only to support individual care, but also quality improvement initiatives as well as accelerating knowledge generation, e.g. through the use of comparative effectiveness research methods [10].

The target patient groups

The burden of severe mental illnesses (SMI) like schizophrenia, bipolar disorders, and depression, continues to grow, with significant impact on health and major economic consequences in all countries of the world [11]. Individuals with severe mental illness are also disproportionately affected by comorbid somatic conditions [12], resulting in a life expectancy that is about 20 years shorter than the rest of the population, in the OECD countries [13]. There is room for improvement. Even in countries in the top in OECD healthcare quality rankings, like Sweden [14], 71.7% of 29823 patients in the nationwide cohort with schizophrenia spectrum disorders experienced treatment failure, defined as discontinuation or switch to other medication, suicide attempt or death [15]. The latest decades have seen the rise of potential game changers, like the LHS, in other domains of healthcare.

LHS in mental health services

The number of reports, books and peer-reviewed publications related to LHSs have grown rapidly, and real-world applications are starting to present promising results [16], and illuminating facilitators and barriers [17,18]. However, what about mental health and severe mental illness?

When performing a review of the literature on LHSs, almost no papers related to mental health services and SMI were found [19]. Only one publication had a description of a mental health-related LHS implementation [20]. This was done in scaling up a dementia and depression collaborative care model from initially 200, up to 2000 older adults, within 24 months. Some data on improvement are presented: hospital length of stay, readmissions, reduction in mortality and reduction in central line-associated bloodstream infection rate. However, there is no explanation on how this was related to the development of a learning health system. Nor are there any descriptive statistics on how the initial group of 200 patients differed from the 2000 that later were included in the collaborative care model.

Similarly only one publication presented conceptual contributions specifically for mental health and SMI. Green et al. [21] aimed at providing a framework needed to achieve some overarching objectives related to the development of mental health as a learning health system, supporting the use of relevant outcome measures and the use of comparative effectiveness research (CER) on service structure and delivery. It is an extensive article with a broad scope and much detail, in summary implicating: (1) a lack of sufficient outcome measures that measure the things that are most important for the service users, since more value-based information is fundamental to provide person-centered care, (2) a need to establish links between person-centered outcome measures and processes, interventions and structures to provide feedback on what works where, when and for who, and (3) that outcome data should be used to support care for the individual patient, guide management and be sufficient in conducting CER. Green et al. [21] conclude that collaborations are needed to address the fragmentation of care due to the tension between different stakeholders and their needs.

Hatch et al. [22] concluded that there are many facilitators and barriers related to patient characteristics and potential motivators when implementing digital health tools for patients with SMI. Therefore, the objective of this paper is to explore, in general, how individuals with SMI potentially could benefit from an LHS4SMI.

User-centered system design

Digitalization of healthcare requires ‘a profound understanding of the users and their context and the entire health-care system and thus requires a user-centered systems design approach.’ [23,p.72]. User-centered system design (UCSD) [24] is getting more common [25] and has been the method of this study, using key principles in providing examples of what an LHS for people with SMI could be experienced like. A composite of real-life cases, a persona (Ana), was created by authors AG, UM, and LL. Use-case scenarios [26] were chosen to help highlight essential differences between a non-LHS and an LHS.

The Persona: Ana

Ana, a 34-year-old woman, has suffered from psychotic symptoms since she was in her early twenties. The early symptoms were primarily depression and anxiety. Her parents had separated earlier, and her father moved out of town. She had exceptional grades in school and enjoyed studying, but was forced to quit studying at the university due to increased stress, some of it possibly a result of cognitive impairment and decreased social functioning. When hallucinations and delusions increased and were obvious, she was put on antipsychotics for the first time. She could not cope with living on her own, being depressed and sometimes suicidal, moved from her apartment and stayed at her mother’s place. She consulted with several psychiatrists over the following years. Several diagnostic procedures resulted in a schizophrenia diagnosis at age 28. Different drug treatments were attempted but resulted in side effects hard to cope with. Clozapine was introduced and is the primary antipsychotic drug she currently uses. Her drug treatment is combined with various psychosocial interventions. Three psychotic episodes have, since the onset of her psychotic symptoms, resulted in emergency inpatient care at the local hospital. Even though she fears another episode, she longs for her apartment and the possibility to continue her studies at the university. A year has passed since she last met with her psychiatrist for a yearly check-up and she is unsure on whether her medication is effective, or if there is anything else that can be done to increase her chances of living independently, creating her own life. It worries her.

We will take a brief closer look at three different points of time in a use-case scenario: (1) Ana at a yearly check-up meeting with a psychiatrist, (2) Ana reflecting on the visit when going back home by bus, and (3) Ana in crisis, visiting the emergency room at the hospital. These will be described in two different ways, one in a non-LHS, and the other, as Ana is getting care within an LHS.

Non-LHS visit to the psychiatrist

Ana does not know how to answer the questions the psychiatrist asks her. Was she feeling better? Had she noticed any increase in side-effects? She became aware of her so-called residual symptoms when the voices returned, but that did not bother her much right now. A possible change of medication worried her more. The psychiatrist seemed indecisive as if trying to decide which alternative would be better: increasing the dosage of her current medication, or switching to another one. Ana has tried different medications over the years. When seeing different psychiatrists, they all seemed to want to improve the treatment by switching medication or changing the dose. This time, the psychiatrist proposes not to make any changes in the current medication, and Ana agrees to continue as before.

LHS visit to the psychiatrist

Ana is used to the yearly procedure of filling out questionnaires and then talking briefly about important aspects of life, with a deeper probing into areas of uncertainty. Ana knows that the psychiatrist makes a general assessment while assessing symptoms related to Ana’s illness. He is ticking off boxes on a tablet. Ana is aware of her diagnosis. Her knowledge of schizophrenia has increased over the years, not only thanks to her outpatient clinic but also websites and patient forums. Ana and the psychiatrist looked at a digital visualization of Ana’s care plan and progress. She could see how her symptoms and level of functioning had changed over time, a confirmation of things moving in the right direction. By using the clinical decision support system, prognostic information on symptom reduction and side effects of different treatment options are adjusted to the characteristics of Ana’s illness. Ana and the psychiatrist looked thoroughly and discussed various options before agreeing on a care plan.

Non-LHS reflections after the visit

Ana is on the bus on her way back home. She knows that her mother is going to ask her later about the visit to the psychiatrist. Her mother always worries about her. Ana is planning to tell her that she will continue the same medicine as before and the same dosage. She will also tell her mother that the goal of the care plan is for her to get well enough so that she can resume her studies. That might calm her a bit even if Ana just does not know how it could be possible.

LHS reflections after the visit

During the bus ride home, Ana starts thinking about how things have evolved. Even though she is aware of her problems and challenges, she is also encouraged by her progress. Using her smartphone, she logs into the patient portal to take another look at the graphs showing the improvement of symptoms and level of functioning in the latest years. She feels hopeful about the future, especially after today’s visit. She writes an instant message in the portal that can be read by her trusted friends and family in her social support network group: ‘Just been at the clinic. Things are going well. Hoping to get back to university this fall!’ She gets a couple of messages back within minutes. Among them, one from her mother: ‘Good news! We are here for you!’

Non-LHS visit to the emergency room at the hospital

Ana barely found her way to the emergency room at the hospital. Stress and insomnia have triggered psychotic symptoms, and now she is desperate to get help. She has a hard time explaining what is wrong and has to sit and wait among the other late night visitors to the emergency room. It is a frightening experience. When finally seeing a psychiatrist, it is someone whom she has never met before, and he seems to have difficulties finding relevant information in the electronic health record to support him with clues on how to handle the situation — Ana’s anxiety spikes. She feels alienated and fears that she will not get the help she expected. She fears that she will be committed when the psychiatrist calls a colleague to discuss which course of action to take. It feels as if she has no say about admission to the ward. It took almost four months before she got back home, the last time she was at the ward. How long will it be this time?

LHS visit to the emergency room at the hospital

Stress and insomnia have triggered psychotic symptoms, and Ana found her way to the emergency room at the hospital for help. She has a hard time explaining what is wrong but remembers to refer to her action plan which she earlier had set up with her case manager. It lists her ‘early warning signs’ and her chosen ‘strategies’, as well as information on how she wishes to be received and treated in a situation like this. Even though being anxious, she finds it comforting when the psychiatrist, who she never met before, reads the action plan out loud, asking Ana to confirm that the information is correct. He asks her if she would like him to call her mother since that has the highest priority in her action plan in times of crisis. Ana nods. It takes some time before her mother arrives. She knows about Ana’s action plan and how Ana would like to be supported when in crisis. Ana, her mother, and the psychiatrist discuss the situation. Is it even necessary to admit Ana to the inpatient ward? Ana wants to return home with her mother, now knowing what to do if sleeping pills do not help and her situation does not improve. Relevant information is documented for her Case Manager to access first thing in the morning when at work, thus enabling mobilization of support for Ana.

An LHS for patients like Ana

The case of Ana getting care in an LHS proposes several advantages compared to the non-LHS scenario. To be listened to and participate in all the decision making related to treatment are experienced as pivotal in the real-life cases used to create the persona of Ana. The LHS-scenarios illustrate how care processes can be supported by visualizations of information in an LHS, involving the patient in the coproduction of care, through the application of shared decision making [4,5]. As stated by McGinnis et al. [1] a well working LHS is seamlessly embedded in the patients’ care processes. The LHS, generating useful visualizations of information when needed, can serve as an ‘external memory’, keeping information on progress and earlier treatment attempts as well as enabling participation even in times of crisis and cognitive impairment, increasing patient safety. Continuous evaluation of treatment through the use of patient-reported outcome measures [6] can, along with psychoeducation [27], support increased understanding of how to handle symptoms and early warning signs [28].

The inclusion of next of kin is stressed both in the definition of what an LHS is McGinnis et al. [1] as well as in treatment programs based on clinical microsystems for service delivery [29]. Patient portals and communication platforms can support self-management of illness, psychoeducation and the inclusion of next of kin, in the evaluation of treatment and in action plans.

An LHS can enable learning on all levels of a healthcare system. Joint learning and increased understanding can be obtained by the use of structured follow-ups, measuring what is important for the individual, and using the harnessed knowledge obtained from every previous treatment of similar patients. The access to all relevant data for all patients in certain patient groups can potentially enable faster diagnostic processes, support better individualized real-time guidelines as well as in supporting quality improvement initiatives and comparative effectiveness research [10,21]. This has the potential to shorten the time from debut to diagnosis as well as finding the most effective treatment faster, reducing suffering, potential harm and costs for both the individual and the healthcare system.

Conclusions

The case scenarios illuminate the differences between a healthcare system working as an LHS and not, proposing, that an LHS4SMIs like schizophrenia has enormous potential in enabling continuous learning, patient coproduction and more effective care for someone like Ana. Continued development and research are needed to find paths forward towards a Learning Health System for Severe Mental Illness (LHS4SMI).

Acknowledgments

The authors would like to thank Peter Jacobsson and Ulla Karilampi for comments on a late draft.

Disclosure statement

The authors declare that they have no competing interests.

Additional information

Funding

The study was financially supported by the Swedish Schizophrenia Fellowship.

Notes on contributors

Andreas Gremyr

Andreas Gremyr is a Psychologist and Quality Improvement Officer, at the Department of Schizophrenia Spectrum Disorders at Sahlgrenska University Hospital. He is conducting doctoral studies related to the field of learning health systems and severe mental illness at the Jönköping Academy for Improvement of Health and Welfare, Jönköping University.

Ulf Malm

Ulf Malm is an Associate Professor at Gothenburg University with extensive experiences as a clinical psychiatrist and researching mental health services. His primary focus of interest has been outcome research focused on drug therapies, psychotherapies and psychosocial interventions, evidence-based treatment and care of schizophrenia, assessment methods, and service delivery by case management and clinical microsystems for people with severe mental illness.

Lennart Lundin

Lennart Lundin is the first Vice President of the Swedish Schizophrenia Fellowship, a patient advocacy association supporting individuals and families affected by schizophrenia spectrum disorders as well as working towards increased knowledge for politicians, policy- and decision makers in health and welfare.

Ann-Christine Andersson

Ann-Christine Andersson is an Associate Professor in Quality Improvement and Leadership in Heatlh and Welfare at the Jönköping Academy for Improvement of Health and Welfare, Jönköping University. Her main areas of research is within Improvement Sciences and Co-production.